IoT botnets powered by Mirai continue to grow
- 19 October, 2016 22:57
Level 3 Threat Research has noted an uptick in activity by new IoT botnets that are backed by the Mirai malware, with some attacks enlisting 100,000 individual hijacked devices.
A significant number of these zombie devices are enslaved by more than one botnet, according to the research described in the Level 3 Beyond Bandwidth blog, and some of these botnets use overlapping infrastructure.
Source code for Mirai was released Sept. 30, “which has inspired a significant number of new bad actors, all working to exploit similar pools of vulnerable devices,” the Level 3 researchers write.
They came across Mirai in mid-September, just days before it gained notoriety when it was used to launch one of the largest ever DDoS attacks, taking down the Krebs on Security Web site.
Before that, they had been tracking a different IoT botnet malware called BASHLITE, which is still around and being used to run giant IoT botnets in competition with owners of Mirai-enabled IoT botnets. Level 3 continues to monitor both, which are used to support DDoS-for-hire businesses.
Mirai is more sophisticated than BASHLITE, says Level 3 CTO Dale Drew, and since Mirai has been released, it is becoming still more sophisticated. For example, Mirai owners now rotate the IP addresses of the botnet command and control (C2) servers more frequently, every day or so, to avoid detection. That’s about three times as often as has been observed with other IoT botnets, the research says.
Actors using MIrai sometimes use the same IP addresses for C2s but assign different domain names to them over time. Those C2s are mostly compromised servers in small and midsize businesses, Drew says, and Level 3 is reaching out to the victimized organizations so they can clean up their machines.
Two of the Mirai C servers were attacked simultaneously at gigabit-per-second levels over a period of 24 hours stretching from Sept. 18 to Sept. 19. The attacker was BASHLITE. Three shorter, lower bandwidth attacks were subsequently noted on separate Mirai command and control servers on Sept. 20, 22 and 23.
Those shorter attacks might have been attempts to take over control of the C2s to enlist them and the bots they controlled into BASHLITE, Drew says.
Since the release of Mirai source code, the geographic distribution of hijacked IoT bots has changed. The code contains common and default user names and password, and these dictionaries have made attacks more effective in the U.S., so now a higher percentage of devices than with BAHSLITE are located here. The most are in the U.S. (29%), then Brazil (23%) and Colombia (8%).
Level 3 has tracked the IP addresses that contact known command and control servers for both kinds of botnets, and track what devices reach out to them. Prior to Mirai code being released, Level 3 had discovered 213,000 IoT devices being used as bots. Afterwards, that number more than doubled to 493,000. Since Level 3 doesn’t have a complete view of the bot infrastructure, the actual number is likely higher, the researchers say.
The research found that individual hijacked IoT devices were used to scan the internet for other vulnerable IoT devices on TCP ports 23 and 2323 and send the IP addresses and credentials of vulnerable devices to a reporting server. Loader servers would then connect to the devices that had been designated as vulnerable and downloaded loaders to them. Then the infected machines connected to separate servers to download Mirai.
Level 3 observed long-lasting connections between the reporting servers and the onion router (Tor) network, which the researchers concluded was the bot herder connecting to control the botnet.
More than 80% of the bots were DVRs, with the rest being IoT routers, cameras, Linux servers and the like using default user names and passwords.
The researchers found that 24% of the bots in the Mirai botnets had also been enslaved by BASHLITE. “Such a high overlap indicates that multiple malware families are targeting the same pool of vulnerable IoT devices,” they say.